Audio files and R code supporting the analysis of vessel noise and Opsanus tau calling behavior in New York City

Quigley, Thomas 1 2 ; Lazrinth, Xylo1; Gdanski, Sydney1; Rodriguez, Rachel1; Forlano, Paul1 2

Published Jun 29, 2026 on Dryad. https://doi.org/10.5061/dryad.xgxd254v2

Data files

Jun 29, 2026 version files 10.78 GB

20_min.zip

1.18 GB
call_behavior_bda_noise.R

19.72 KB
combined_data.csv

160.45 KB
noise_call_characterization_code.R

24.47 KB
README.md

8.55 KB
Slice_1.zip

9.60 GB

Abstract

The purpose of this data was to analyze the impact of vessel noise on the calling behavior of a soniferous fish. Female oyster toadfish (Opsanus tau) must be able to hear the male advertisement call in order to reproduce, but vessel noise occurs at frequencies that can mask those calls. New York City provides reproductive habitat for toadfish and also has a high level of vessel traffic. New York City is, therefore, an ideal location to study the effects of noise on fish. We used the data to examine whether toadfish change their calling behavior in the presence of vessel noise.

The submission includes audio files in .WAV format, recorded underwater along the coast of New York City in May and June 2023. Also included is a .csv summary of oyster toadfish advertisement calls identified in the audio files. Also, two R code files that analyze this call data are included. The R code uses linear models to examine the impact of several parameters of vessel noise (incidence, amplitude, dominant frequency, duration, and periodicity) on toadfish calls. Linear models also examine several parameters of calls (number, amplitude, fundamental frequency, duration, and periodicity) and relate them to incidence of vessel noise. Finally, for some analyses, Raven Pro selection tables that annotate toadfish calls in .WAV files are included.

The audio files include other biophony including sounds of the striped cusk eel (Ophidion marginatum), white perch (Morone americana) and other organisms that we did not identify. The files could be re-used to study other estuarine species and their behavior in noise.

Dataset DOI: 10.5061/dryad.xgxd254v2

Description of the data and file structure

The audio files were the output of a passive acoustic monitoring effort in New York Harbor. The files include 4 days of samples from four locations in the harbor from May and June of 2023. All locations are within the city of New York. The locations were:

Name	Abbreviation	Lat	Lon
Pier 40	p40	40.728272	−74.011956
Valentino Pier	val	40.678267	−74.018401
Sheepshead Bay	shb	40.583017	−73.936247
Shirley Chisholm State Park	csp	40.638069	−73.868471

Files and variables

The Raven selection tables trace oyster toadfish advertisement calls in the audio files, and include other metadata on the calls. The selection tables were used to generate combined_data.csv, which summarizes the data. The R files analyze the data either from combined_data.csv or from selection tables directly.

File: noise_call_characterization_code.R

Description: Characterizes vessel noise and calls and examines the effect of noise on call number.

File: call_behavior_bda_noise.R

Description: Further examines the effect of vessel noise on calls, especially on call amplitude, frequency, and duration.

File: combined_data.csv

Description: A summary of the data derived from the Raven selection tables. Each row summarizes one sample and each sample is a 4-minute recording.

Variables

date: The date of the recording.
hour: The hour (0-23) of the recording.
minute: The exact start minute of the recording (either 00 or 30).
calls: Number of calls identified in the 4 min sample, using the R library ohun for call detection.
date_time: A convenience field that concatenates date and time.
location: One of four experimental locations (val, p40, shb, csp).
period: The period of the day in which the recording was made. Periods were night (0-6 hrs), morning (6-12), afternoon (12 - 18), and evening (18 - 0 hrs).
calls_per_min: The number of toadfish advertisement calls divided by the length of the sample in minutes.
hour_prop_daily_calls: The proportion of the entire day's calls that occur in this sample. NAs indicate that the value is unknown because not all of that calendar day's hours were sampled.
duration: Duration of sample, in seconds.
non_ot_rms_fs: Several subsamples that did not include toadfish calls (non_ot) were examined to estimate amplitude (rms) of all other sound, in this case in the full possible frequency range (full scale or fs) of the recording (up to 24 kHz). Units are decibels.
non_ot_rms_hr: Several subsamples that did not include toadfish calls (non_ot) were examined to estimate amplitude (rms) of all other sound, in this case in the hearing range (hr) of toadfish (up to 800 Hz). Units are decibels.
ambient_rms_fs: Several subsamples with only ambient sound (ambient) were examined to estimate amplitude (rms) of ambient sound, in this case in the full possible frequency range (full scale or fs) of the recording (up to 24 kHz). Units are decibels.
ambient_rms_hr: Several subsamples with only ambient sound (ambient) were examined to estimate amplitude (rms) of ambient sound, in this case in the hearing range (hr) of toadfish (up to 800 Hz). Units are decibels.
vessel_rms_fs: Maximum amplitude of vessel noise in the sample, in the range up to 24 kHz. Units are decibels. NAs indicate no vessel noise was identified in the sample.
vessel_rms_hr: Maximum amplitude of vessel noise in the sample, in the range up to 800 Hz. Units are decibels. NAs indicate no vessel noise was identified in the sample.
vessel_dominant_fs: Dominant frequency of vessel noise in the sample, in the range up to 24 kHz. Units are hertz. NAs indicate no vessel noise was identified in the sample.
vessel_dominant_hr: Dominant frequency of vessel noise in the sample, in the range up to 800 Hz. Units are hertz. NAs indicate no vessel noise was identified in the sample.
vessel_duration: Duration of vessel noise in sample, in seconds. NAs indicate no vessel noise was identified in the sample.
vessel_proportion_of_sample: Proportion of sample occupied by vessel noise. E.g., 200 seconds of vessel noise in a sample would be 83.7% of the sample or 0.8368.
period_prop_daily_calls: The proportion of the entire day's calls that occur in this sample's period. NAs indicate that the value is unknown because not all of that calendar day's hours were sampled.
censor: Whether the row was censored from analysis because of too much overlap from vessel or other noise.
call_fundamental_fs: Average advertisement call fundamental frequency, up to 24 kHz. Units in hertz. NA indicates that too few calls were recorded to calculate frequency.
temperature: Water temperature in C at the time of the sample.
noise_detected: Boolean indicating whether or not noise was detected in the sample.
freq_overlap_fundamental: Boolean, T if the dominant frequency of vessel noise overlapped the average fundamental frequency of calls. NA for sample without noise.
freq_overlap_2nd_harmonic: Boolean, T if the dominant frequency of vessel noise overlapped the average second harmonic of calls. NA for sample without noise.
freq_overlap_3rd_harmonic: Boolean, T if the dominant frequency of vessel noise overlapped the average third harmonic of calls. NA for sample without noise.
call_amplitude: Average call amplitude in sample in decibels.
calls_in_total_quiet: Count of calls in sample if there is no noise.
calls_in_quiet: Count of calls in sample that did not overlap noise, in sample that includes noise.
calls_in_noise: Count of calls in sample that overlap noise, in sample that includes noise.

File: Slice_1.zip

Description: Contains 4-minute sound files and associated selection tables, organized in subfolders by sampling location. These sound files and selection tables are used in noise_call_characterization_code.R. Selection tables include one row for each advertisement call / noise event that was identified. Selection tables include some columns necessary to open the files in RavenPro.

Variables

Selection: The unique ID of the selection (call or noise) in the file.
View: RavenPro view (can be ignored).
Channel: RavenPro source channel (can be ignored).
Begin Time (s): Start point of the trace, in seconds.
End Time (s): End point of the trace, in seconds.
Begin File: Wav file that this trace applies to.
Delta Time (s): Length in s of the trace.
template: Toadfish call template, applied via the R package ohun, used to identify the call.
scores: Score that ohun gave to the template.
Dominant: Dominant frequency of the trace, in hertz.
Amplitude: Amplitude of the trace in decibels.
Fundamental: Fundamental frequency of the trace, in hertz.
call_occurs_in_noise: Boolean indicating if this trace occurred in noise.

File: 20_min.zip

Description: Contains several 20-minute sound files and associated selection tables, organized in subfolders by sampling location. These sound files and selection tables are used in call_behavior_bda_noise.R. Selection tables include one row for each advertisement call / noise event that was identified. If a variable is not described below, its definition was included above under Slice_1.zip.